Occasionally, I see something mentioned about tracing a particular cyrus process to determine what's going on (or wrong, as the case may be). How is this done? I'm not getting very far at resolving my particular problem, so I'd like to know what the proxyd process thinks it's doing each time it spawns a new connection to itself.

Rob

On 7/1/11 9:10 AM, Robert Spellman wrote:
We have recently upgraded to cyrus 2.4.6. Our environment includes a server running as a murder server, four back end mailstores, and two front end servers. If we mistakenly create a user mailbox on one of the front end servers using cyradm, and then try to manage it (dm, sam, info, lm), cpu load on the front end server increases until the box is unusable. Syslogs show the user cyrus (our admin user) logging in over and over, each time creating a new proxyd process, which explains the high cpu load.

Managing mailboxes on the back end servers from the front end servers works fine.

Here's my imapd.conf for one of my front end servers:

configdirectory: /var/lib/imap
partition-default: /home/imap
defaultpartition: default
sievedir: /var/lib/imap/sieve
admins: cyrus backend
sasl_pwcheck_method: saslauthd
sasl_mech_list: PLAIN GSSAPI

altnamespace: true
hashimapspool: true
allowallsubscribe: true
allowusermoves: true
autocreateinboxfolders: Trash|Sent|Junk
autosubscribeinboxfolders: Trash|Sent|Junk
autosubscribesharedfolders: shared.announce
lmtp_downcase_rcpt: true

tls_cert_file: /etc/pki/tls/certs/mail.bates.edu-cert.pem
tls_key_file: /etc/pki/tls/certs/mail.bates.edu-key.pem
tls_ca_file: /etc/pki/tls/certs/bates.edu-cert.pem

proxy_authname: backend
proxyservers: backend

mupdate_server: murder.bates.edu
mupdate_username: backend


----
Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/

<<attachment: rspell.vcf>>

----
Cyrus Home Page: http://www.cyrusimap.org/
List Archives/Info: http://lists.andrew.cmu.edu/pipermail/info-cyrus/

Reply via email to